Abstract
Vision based localization is widely investigated for the autonomous navigation and robotics. One of the basic steps of vision based localization is the extraction of interest points in images that are captured by the embedded camera. In this paper, SIFT and SURF extractors were chosen to evaluate their performance in localization. Four street view image sequences captured by a mobile mapping system, were used for the evaluation and both SIFT and SURF were tested on different image scales. Besides, the impact of the interest point distribution was also studied. We evaluated the performances from for aspects: repeatability, precision, accuracy and runtime. The local bundle adjustment method was applied to refine the pose parameters and the 3D coordinates of tie points. According to the results of our experiments, SIFT was more reliable than SURF. Apart from this, both the accuracy and the efficiency of localization can be improved if the distribution of feature points are well constrained for SIFT.
Highlights
Localisation with Global Navigation Satellite Systems (GNSS) in dense urban areas suffers from masks of signals and multi-path errors and leads to significant errors
Different with Global Bundle Adjustment (GBA) which aims to minimize the sum of squared backprojection errors (Triggs et al, 2000), Local Bundle Adjustment (LBA) will minimize the back-projection errors under the constraint of prior poses Cp0 in each sliding window, the cost function is shown in equation 3
It is obvious that the time spent on feature extraction and matching is much higher than LBA in our localization procedure from figure 6e, where the blue and green bins are much higher than the red bins
Summary
Localisation with Global Navigation Satellite Systems (GNSS) in dense urban areas suffers from masks of signals and multi-path errors and leads to significant errors. The local feature based methods have been proved to be an excellent choice for pose estimation. The factors such as illumination changes, perspective deformations and moving objects make the matching a difficult task and influence the accuracy of pose estimation. The SIFT (Scale Invariant Feature Transform) which is invariant to scale change , rotation and illumination (Lowe, 2004) has been applied in the vision based localization (Se et al, 2001; Yang et al, 2009). The SURF is robust to the change of scale, orientation and illumination and is used for feature extraction in pose estimation methods (Murillo et al, 2007). Introduces the SIFT, SURF, grid based extraction and the matching method for localization.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have